Supporting and understanding the health of patients with chronic diseases remains a major challenge. Visualization has the potential to provide personalized and person-centered care [1]. The health data generated are often provided as feedback to patients, and visualization plays an important role in facilitating the interpretation and understanding of the data and, therefore, influencing their actions [2]. Visualization is being used to show patient outcomes in an increasing number of studies [3]. In addition, the review by Ooge et al [4] points to a lack of web-based visualization monitoring systems and systems aimed at laypeople. Visualization, such as pictures, sketches, charts, graphs, and diagrams, can help communicate health information usefully. Visualization can simplify the presentation of complex information and make it more appealing [5]. Algorithmic outcomes can typically be visualized in different ways, depending on the algorithm and the insights being sought. These insights are usually connected to health care activities, which more often focus on interpreting data rather than predicting or monitoring them [4]. These insights can be used to support both written and spoken health messages [5].

Digital health tools can help people with estimation of cardiovascular disease (CVD) risk by empowering or encouraging them to adopt healthier lifestyle habits using different techniques such as visualization to support actionable insights. This is a key public health strategy to prevent or treat CVDs [6]. In a study in which people were randomly selected, the Framingham Risk Profile (GFRP) decreased after 1 year for participants who saw visual imaging results and increased for the group that only saw the risk scores [7]. Some risk communication strategies such as percentages, bar graphs, and icon arrays, which provide patients with a probability, fail to increase risk perception [8,9]. Many of the most frequently used CVD risk scores, such as the GFRP, consider a 20% “risk of developing CVDs in the next 10 years” to be high. Because 20% appears in the lower part of the graph, these scores can be interpreted as low risk. The same is true for icon arrays, where many positive icons make it easy for patients to believe that they are unaffected [10,11]. CVD health assessment feedback is a method of presenting personalized risk information [12]. Providing additional evidence on CVD risk to individuals, such as that shown on heart scans or with a heart age above the individuals’ actual age, may provide a cue to action [13-15]. This is consistent with previous research where strategies using imaging or visualization were most useful in communicating personalized risk [16-18].

Several studies have demonstrated the potential of visualization tools not only in the estimation of CVD risk but also in influencing patient behavior. Turchioe et al [2] found that a line chart was the most used, particularly for data collected over a longer period. They found that patients had a better understanding of line graphs and bar graphs and that color effectively conveys risk, enhances comprehension, influences patient behavior, and boosts confidence in interpretation. Backonja et al [1] found that the use of colors and reference lines was helpful in interpreting data, which subsequently motivated patients to make healthier lifestyle choices. They also revealed that visualization provides many opportunities for explainable artificial intelligence in health care by providing insights into advanced algorithms through visualization, interaction, guidance, and direct explanations [4]. These findings underscore the importance of effective visualization in not only informing patients about their health status but also motivating them to take actionable steps to reduce their risk.

This review explored the potential benefits of visual interpretation for patients with CVDs, the world’s leading cause of death. Specifically, it aimed to explore how visualization techniques can influence patients’ understanding of their risk and motivate them to adopt healthier behaviors. This review focused on the impact of visual aids on risk perception and whether they lead to significant changes in lifestyle or treatment adherence in patients with CVDs.

The PubMed, CINAHL Ultimate (EBSCO), MEDLINE, and Web of Science databases were searched. The search also included gray literature from Google Scholar, where we did not review all the articles, only the highest-ranked ones, and included them according to the relevance of their content. We used the following search string: (“visualization” OR “visualisation tool*” OR “visual interpretation” OR “visual analytic*” OR “visualisation intervention*” OR “chart*” OR “data visualisation” OR “visualisation techniques” OR “visual representation”) AND (“cardiovascular disease* risk” OR “heart disease* risk” OR “cardiac disease* risk” OR “vascular disease* risk” OR “coronary heart disease* risk” OR “CVD risk”) (Textbox 1).

The review included articles published in English, the population included patients and research focusing on estimation of CVD risk, and the comparisons included different types of visualizations (related to digital health, mobile health, apps, images, charts, decision support systems, and other types of visualizations) for estimation of CVD risk. Only empirical studies were included.

Studies that did not involve patients or content about estimation of CVD risk or comparisons related to visualizations were excluded. Studies such as commentaries, editorials, and systematic and scoping reviews were excluded. We also excluded articles that were irrelevant and did not focus on the area under review (Textbox 2).

The search string retrieved in 14 results in PubMed, 495 in CINAHL Ultimate and MEDLINE, 265 in Web of Science and 2 in Google Scholar. In total, 2 authors analyzed the articles using the computer program Rayyan (Rayyan Systems Inc) [20]. Duplicate articles were removed before assessing their eligibility based on their titles and abstracts. If there was disagreement between the authors, a third author was consulted. The articles that passed this evaluation stage went through full-text analysis. We used the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart [21] to describe the review process. In addition, 2 authors individually used the extraction algorithms using the standardized Joanna Briggs Institute data extraction tool [19] (Multimedia Appendix 1).

First, we identified 774 records in the databases. After removing duplicates (230/774, 29.7%), we excluded records that were not in English (102/544, 18.8%) and had inappropriate titles (142/544, 26.1%) and abstracts (140/544, 25.7%). Then we get the reports (160/544, 29.4%) and records that could not be retrieved (62/160, 38.8%). In the next step, we excluded reports with inappropriate content (not focused on CVD prevention; 25/96, 26%), inappropriate study types (protocols; 23/96, 24%), and inappropriate study populations (children; 35/96, 36%). In addition, we reviewed only the highest-ranked results on Google Scholar, and we obtained 2 hits, which we included in the final analysis. A total of 17 studies were included in a scoping review (Figure 1 [22]).

Of the 17 identified studies, the most were from India (n=4, 24%), followed by the United States (n=4, 24%), Iran, Italy, and the United Kingdom (n=2, 12% each). Single studies were also identified by the authors from Oman, Australia, and Spain (each: 1/17, 6%). Descriptive study—model development (4/17, 24%) was the most used methodology, whereas quantitative studies and population-based longitudinal studies (1/17, 6%) were the least used methodology. The largest number of participants was found in the study by Bonner et al [13], which included 361,044 participants who used a heart age calculator. The study developed and validated a web-based heart age calculator. The smallest number of participants (N=70) was identified in the study by Fadel et al [23], which was an experimental study using visual analytics with a dashboard. All visualization methods were based on prognostic models for estimation of CVD risk.

The most used prognostic risk factors were age, sex, and blood pressure (16/17, 94%); smoking status (14/17, 82%); diabetes status (11/17, 65%); family history (10/17, 59%); high-density lipoprotein and total cholesterol (9/17, 53%); and triglycerides and low-density lipoprotein cholesterol (6/17, 35%). Other variables were used less frequently as predictors of CVD risk (Table 1).

We compared the results of the 17 studies on many different aspects. All the studies had the common aim of investigating the usefulness and comparability of the tools for estimation of CVD risk in different populations and settings. Most of the studies (12/17, 71%) were conducted among the general population, but some (5/17, 29%) focused on a target population of patients with different diseases (diabetes, rheumatoid arthritis, and hypertension, as well as patients using lipid-lowering therapy). However, they were different in terms of the specific purposes and contexts of their implementation. Some studies (4/17, 24%) focused on comparing ≥2 tools for the estimation of CVD risk [14,24,31,32], whereas others (13/17, 76%) examined the effect of a single tool for the estimation of CVD risk on the behavior, knowledge, decision-making, or quality of care of individuals or groups [13,23,29,37,38].

The studies used different tools and prognostic models to estimate CVD risk based on different factors and parameters. Some prognostic models were simpler and did not require laboratory tests, such as the GFRP [24], World Health Organization (WHO) and International Society of Hypertension (ISH) risk prediction charts [24,28,29,36], Cardiovascular Risk Score (QRISK2), and a simplified Persian atherosclerotic CVD risk stratification (SPARS) [27], whereas others were more complex and required laboratory tests, such as the Joint British Societies recommendations on the prevention of CVD (JBS3) [14], Systematic Coronary Risk Evaluation (SCORE), and Framingham-Registre Gironí del COR (REGICOR) [26]. Some tools and prognostic models were designed for estimation of CVD risk in the short term (eg, 10 years), such as the GFRP, WHO and ISH, QRISK2, SCORE, and REGICOR [14,24,26,39], whereas others were designed for estimation of CVD risk in the long term, such as the JBS3 and SPARS [27]. Some tools presented the estimation of CVD risk as a number (GFRP, WHO and ISH, QRISK2, SCORE, and REGICOR), whereas others as visual elements, such as cardiac age [14,25,27,38] and estimation of CVD risk [26].

Technology-based interventions have been shown to increase the usefulness of tools for the estimation of CVD risk and can affect several outcomes, such as increasing users’ knowledge, perception, motivation, intention, self-efficacy, satisfaction, compliance, and quality of care regarding their CVD risk and suggesting potential actions to reduce it; changing users’ behavior, lifestyle, risk factors, biological parameters, clinical outcomes, and overall CVD risk to obtain better outcomes; and improving clinical staff’s performance, job satisfaction, confidence, communication, decision-making, and quality of care when the estimation of CVD risk tools (Table 2).

Some of the studies we reviewed (12/17, 71%) used technology-based interventions to improve the effect of tools for the estimation of CVD risk on participants’ behavior, knowledge, decision-making, or quality of care. These interventions took the form of charts, tables, and diagrams (9/17, 53%) and apps (3/17, 18%). These interventions had different characteristics such as 1. presentation formats: displaying CVD risk in different formats, such as numbers, colors, and graphs; 2. user interactivity: allowing users to influence the estimation of their CVD risk by entering or modifying their own data, such as blood pressure, cholesterol, smoking status, physical activity, diet, and more. 3. Different types of tools and systems: clinical decision support systems (3/17, 18%), dashboards (2/17, 12%), education tools (3/17, 18%), and web-based tools (4/17, 24%) and software (2/17, 12%); 4. User engagement: providing feedback, advice, encouragement, reminders, goals, plans, support, or guidance to users based on the estimation of their CVD risk and needs; In some of the articles (6/17, 35%), the same authors described multiple different types of visualizations for estimating CVD risk, rather than focusing on just one type. This facilitating communication, collaboration, coordination, or shared decision-making between users and clinical staff or between users and other users. The most used format to display data in the studies was “visual cues” (10/17, 59%), followed by “bar charts” (5/17, 29%) and “graphs” (4/17, 24%; Table 3).

We reviewed and compared the results of 17 studies that investigated the usefulness and comparability of different tools for the estimation of CVD risk. These studies were conducted in different countries and contexts, such as Oman, Australia, the United States, the United Kingdom, Spain, and Iran. The tools for the estimation of CVD risk used in these studies were in the form of equations, tables, graphs, or computer programs, such as GFRP, WHO and ISH, QRISK2, JBS3, SCORE, REGICOR, and SPARS. The results of these studies showed differences and similarities between tools for the estimation of CVD risk in terms of their objectives, methods, criteria, results, and limitations.

Of the studies reviewed, we found that only the dashboards by Fadel et al [23], Gidlow et al [14], and Hassannejad et al [27] included a tool that also allowed for goal setting, where the health care professionals and patients can agree on target goals and then calculate and visualize how the risk will change over time if these goals are met and maintained. The latter contributes to a better understanding of the impact of lifestyle or treatment adherence. In addition, Mendez et al [40] suggest that these interactions in the visualization tools themselves can help inform patients about the estimation of CVD risk and improve patient understanding of risk and the potential impact of risk-reducing interventions, which we believe can help patients make more informed and empowered decisions to achieve greater risk reduction.

We also found studies that included cardiac imaging as an additional indicator of CVD risk [41-43]. A randomized controlled trial by Whitmore et al [44] with a sample of 7000 patients showed that cardiac imaging not only helped with CVD diagnosis and estimation of CVD risk but also, importantly, helped educate and motivate people to engage in risk modification or lifestyle changes. This highlights the critical role of diagnostic tools not only in clinical decision-making but also in improving patient compliance with treatment and promoting sustainable lifestyle changes that are essential for long-term CVD health outcomes. An analysis of the visualization techniques used in the different studies showed that the most used visualization type was color plots. Colors are important because, together with warning words, they can attract more attention from users [45]. Color coding in matrices and graphs usually reflects the level of risk [46]. This can be particularly valuable in medical settings, where visual cues can improve patients’ understanding of their health risks, potentially leading to better adherence to treatment recommendations and lifestyle changes. In addition, effective visualizations can simplify complex data, making them more accessible to a wider audience, including patients with varying levels of health literacy. On the other hand, it is important to bear in mind that some patients may have color vision impairment, which may affect the interpretation of the data. It is also important to consider the diversity of cultural backgrounds as colors have different meanings in different environments [47].

Bar charts were the second most common type of visualization, followed by graphs. Less commonly used were special graphs and line graphs. On the other hand, Cates plots, pie charts, level charts, timelines, 3D models, infographics, and matrices were used only once each. This finding indicates the predominance of colored visuals in the presentation of data, which may help improve readers’ understanding and perception of the information. In contrast, a study by van Weert et al [48] found that most participants preferred to see risk in the form of hourly, pie, or bar charts. They also found that younger age, higher mathematical ability, and higher graphical literacy contribute to higher knowledge and understanding of risk scores. This suggests that, while certain visual formats may be more appealing or accessible to the general population, individual differences in cognitive abilities, such as numeracy and familiarity with graphical representations, play an important role in the effectiveness of these visual aids. Therefore, tailoring risk communication to the abilities of the user may enhance understanding and improve decision-making regarding health interventions and risk management.

We found that tools for the estimation of CVD risk can be useful for a variety of purposes and contexts, but they also have some limitations that need to be considered when using and interpreting them. One of the limitations of this paper is that we did not include machine learning classification approaches, which offer important advantages in predicting and classifying outcomes but whose limitations should be considered. Future research should aim to address these limitations by incorporating diverse datasets and using methods that increase the transparency and interpretability of models. We recommend that the selection of tools for the estimation of CVD risk should consider the characteristics of the target population, the availability and quality of the data, the way in which risk is presented, the interaction between users and the tools, and other factors that may affect the tools’ performance and comparability. We also suggest that tools for the estimation of CVD risk should be regularly updated, validated, and calibrated to ensure their accuracy, reliability, and generalizability. Ongoing advancements in machine learning techniques and data collection methods will contribute to more accurate and reliable risk predictions in the future. We hope that this paper will contribute to a better understanding and use of tools for the estimation of CVD risk in practice and research.

We identified some innovative features of tools, such as goal setting, visualization, and cardiac imaging, that could improve the estimation of CVD risk and user engagement in risk reduction. We conclude that the selection of tools for the estimation of CVD risk should be based on several factors, such as the characteristics of the target population, the availability and quality of the data, the display and interaction with risk, and the performance and comparability of the tools. We also recommend that tools for the estimation of CVD risk should be regularly updated, validated, and calibrated to ensure their accuracy, reliability, and generality. Future research should test visualization tools to determine their potential impact on patients and their usefulness for health care professionals.